Artificial intelligence can make you feel extraordinarily capable.
Give it a difficult problem and possibilities appear within seconds.
Ask it to challenge an idea, and another perspective emerges.
Do you need a headline?
Twenty arrive.
Need a business concept?
It can brainstorm all afternoon without getting tired.
But psychology is beginning to uncover an important question hidden underneath this productivity revolution:
What happens to our belief in our own ability to think creatively?
A study appearing in the August 2026 issue of Acta Psychologica examined generative AI use and employee creative thinking through two very different psychological pathways: creative self-efficacy and problematic AI dependence. The researchers describe generative AI as having both cognitive gains and potential losses, with innovative self-efficacy playing an important role in supporting creativity.
ScienceDirect
This gives us a useful way to think about human AI collaboration.
AI can make your thinking bigger.
But ideally, it shouldn’t make your confidence in your own thinking smaller.
What Is Creative Self-Efficacy?
Self-efficacy refers broadly to our belief that we are capable of successfully performing a particular task.
Creative self-efficacy applies that idea to creativity.
It is the internal belief:
I can come up with something.
Not necessarily immediately.
Not necessarily perfectly.
But eventually.
That belief matters because creativity involves uncertainty.
You don’t know whether the first idea will work.
You experiment.
You fail.
You modify.
You try something strange.
Then, sometimes, something remarkable emerges.
If a person gradually begins believing that useful ideas only arrive after consulting AI, something psychologically important may change.
The tool has moved from supporting confidence to potentially substituting for it.
The Two AI Users
Imagine two entrepreneurs.
Both use generative AI every day.
The first says:
“I have an idea. Help me test it.”
The second says:
“I don’t have any ideas. Tell me what to build.”
On the surface, both are using the same technology.
Psychologically, however, the relationship is different.
The first person maintains intellectual ownership.
AI expands the idea.
Challenges assumptions.
Identifies weaknesses.
Suggests alternatives.
The second person begins by surrendering the starting point.
Do that occasionally, and it may be perfectly useful.
Do it for every creative decision, and something else can happen:
The machine becomes the source of permission to think.
Different AI Use Can Produce Different Creativity
This distinction fits other 2026 psychology research.
A study published in Frontiers in Psychology examined generative AI usage among 381 employees in knowledge-intensive organisations.
Researchers distinguished between exploratory and exploitative AI use.
Exploratory use was more strongly associated with radical creativity—ideas representing larger departures from existing approaches.
Exploitative use was more strongly associated with incremental creativity—improving or refining what already exists.
Frontiers
That’s an important distinction.
AI isn’t simply a creativity switch.
How you use it changes what kind of thinking it supports.
Treat AI Like a Cognitive Gym
Perhaps we should stop thinking about AI purely as a productivity machine.
Think of it as a cognitive gym.
A good gym does not lift the weights for you.
It creates conditions in which you become stronger.
AI can do something similar.
Instead of asking:
“Write my strategy.”
Try:
“Here is my strategy. Find the three weakest assumptions.”
Instead of:
“Give me an idea.”
Try:
“Here are my five ideas. Give me three perspectives that contradict them.”
Instead of:
“Solve this problem.”
Try:
“Don’t solve this yet. Ask me questions that might help me discover a solution.”
Now AI isn’t replacing cognitive effort.
It is creating productive cognitive resistance.
The Difference Between Assistance and Dependence
There is nothing inherently wrong with relying on technology.
Humans already rely on calculators.
Maps.
Search engines.
Spellcheckers.
Databases.
Smartphones.
The question isn’t whether we depend on tools.
Civilisation itself is built on tools.
The more useful question is:
Which human capabilities do we want to preserve even when machines can perform them for us?
Navigation software can tell us where to drive.
But perhaps we still want spatial awareness.
Calculators can solve equations.
But we still teach mathematical reasoning.
AI can generate ideas.
Perhaps we should still practise imagination.
A Practical Rule: Human → AI → Human
A useful creative workflow could be:
Human: Generate the initial question or direction.
AI: Expand possibilities, challenge assumptioResearch: Acta Psychologica — Cognitive Gains and Addictive Losses: Generative AI and Employee Creative Thinking⁠ and analyse alternatives
Human: Judge, transform, combine, and decide.
Then repeat.
That keeps the person cognitively present throughout the process.
It also preserves something essential:
agency.
AI Satisfaction Matters Too
The August Acta Psychologica study also examined satisfaction with AI use as part of the relationship between generative AI and creative thinking. �
ScienceDirect
This reminds us that human-AI interaction isn’t purely technical.
It is psychological.
People develop expectations about their tools.
Trust.
Frustration.
Confidence.
Habits.
Even emotional attachment.
The psychology of AI may therefore become just as important as the engineering of AI.
We are not merely designing intelligent systems.
We are designing relationships between intelligence systems and human minds.
The Goal Isn’t Independence From AI
There is a temptation to turn this debate into another false choice.
AI or human.
Machine creativity or human creativity.
Dependence or complete independence.
That misses the opportunity.
The goal isn’t to prove that humans can work without AI.
The goal is to ensure that working with AI makes human capability larger rather than smaller.
A calculator should help a mathematician explore more difficult mathematics.
A telescope should help an astronomer see further.
And artificial intelligence should help humans ask questions they might never have reached alone.
Final Thoughts
The most valuable AI user of the future may not be the person who can generate the most content.
It may be the person whose relationship with AI continually strengthens their own ability to think.
Use AI.
Challenge it.
Learn from it.
Disagree with it.
Let it surprise you.
But occasionally, close the chat window and ask yourself:
What do I think?
If you still have an answer, the partnership is probably working.

Research: Acta Psychologica — Cognitive Gains and Addictive Losses: Generative AI and Employee Creative Thinking⁠